Signal versus Diagnosis: From Burnout Risk Screening to Care

Dear Editor,

This letter addresses specific interpretive and evidentiary points raised by Van Zyl (2026), published in this journal in response to my commentary in Psychiatry Research (De Beer, 2026). Read in context, my use of the phrase “Twilight Zone” concerned a specific diagnostic uncertainty arising from burnout’s ICD-11 status as an occupational phenomenon: what does an elevated workplace risk score mean? Nonetheless, Van Zyl and I share substantial common ground: severe burnout complaints warrant the attention of health professionals, organisations should decide in advance how elevated scores will be handled, and organisational prevention must not substitute for individual care. I will address three matters in this letter: whether my position separated screening from care, whether a high screening score is diagnostically sufficient, and whether such a score must compel clinical triage.

First, I used the phrase “dual approaches” descriptively; I did not propose “separate silos”, nor establish the “Dual Approach Model” that the commentary attributes to me, and I welcome the opportunity to reiterate my position. My commentary described organisational and clinical approaches as “complementary rather than competing” and assigned workplace screening two functions: identifying modifiable working conditions and identifying employees who may require referral, with diagnosis reserved for clinical assessment by an appropriately qualified health professional rather than questionnaires alone (see also De Beer & Schaufeli, 2025). The distinction charged as a “false dichotomy” is one the proposed model preserves; it states that screening and diagnosis are “different activities requiring different competencies” and that the model “does not collapse screening and diagnosis into a single process.” What the model does add is a more formalised process and a prescriptive account of the transition between screening and care, not a refutation of the line between them. The issue, then, is not whether elevated scores may be labelled at all. They may properly be treated as risk signals requiring response. The question is whether that screening label should be converted into a diagnostic label; that step requires clinical assessment by an appropriately qualified health professional.

Second, a high score is a relevant signal but not a diagnosis. The BAT test manual states that no diagnosis can be made from a self-report questionnaire alone and that an anamnestic and diagnostic interview by a trained clinician is required (Schaufeli et al., 2020). Furthermore, the cutoffs are described as preliminary and to be used tentatively (Schaufeli et al., 2023). The reported AUC, sensitivity, and specificity do not remove this limitation. None gives the probability that a particular employee with a red score is a clinical case; that positive predictive value depends on prevalence. A specificity near 90% means that about 10% of non-cases may score positive, not that only 10% of positive scores are false. In a general workforce, where clinical burnout is less prevalent than in the clinically diagnosed samples used to set the cutoffs, false positives may form a sizeable share of red scores. The score identifies candidates for evaluation; it does not make a determination. Biological markers do not close this gap: the response cited on this point reports that no biomarker or questionnaire is precise enough to serve as a standalone diagnostic tool for clinical burnout, and that the BAT, despite strong psychometric properties, should support rather than replace clinical assessment (Vandenabeele et al., 2025). Moreover, Bianchi and Schonfeld (2025) are cited to argue that under-recognising severe burnout leads to misdiagnosis. Yet their argument cuts in the opposite direction: they stress burnout’s overlap with depression as problematic and warn that liberal diagnostic approaches risk ‘diagnosis creep’ — the pathologising of ordinary variations in stress, fatigue, or motivation. My view remains that employees with severe scores should be referred to an appropriately qualified professional to make a clear determination of health status.

Third, “mandatory triage” admits two readings. If it means that an organisation guarantees a prompt, confidential offer of qualified assessment that employees remain free to decline, our positions are close. If it means that a score compels clinical assessment, questions of agency, consent, confidentiality, proportionality, and scope of practice arise. Mandatory interventions can disrupt responsible organisational practice; these concerns are not solved by relabelling screening as triage. Lerouge (2025) does not analyse employer liability arising from workplace screening data, nor argue that individual screening scores should be treated as diagnostic. The article distinguishes a mental-health-at-work approach from a psychosocial-risks approach and places the latter in the employer’s duty to identify, assess, and control risks arising from the organisation of work. The distinction is important because Lerouge cautions that health-promotion framings can individualise the relationship between health and work rather than focus on the organisation of work as a source of psychosocial risks. A screening result may therefore be relevant evidence that psychosocial risks require assessment and preventive action at source, but that preventive logic does not convert the score itself into a clinical diagnosis.

On balance, Van Zyl’s central reminder is important: measurement without a clear response plan is poor practice. Organisations using identifiable screening should plan and explain in advance how data will be handled, give clear feedback and a timely confidential offer of qualified assessment, and act on hazardous working conditions. The proposed model formalises the transition between screening and care; it does not collapse the distinction. The “Twilight Zone” is navigated by connecting a meaningful risk signal to appropriate care without mistaking the signal for the diagnosis.